AbdullrahmanDownload CV
Back to Research
Published 2026-03-27

I Built a Private AI Agent That Can Chat With Any Excel File Locally and Adapt Automatically

Abdullrahman
6 min read

Abstract

A project article about turning Excel and CSV files into a local analytics assistant that adapts to uploaded schemas, answers plain-English questions, and keeps sensitive data on the user's machine.

Project Context

The article describes a local analytics workflow for uploading spreadsheets, asking natural-language questions, and receiving answers grounded in the file data.

It explains how the assistant combines SQL, pandas/Python, semantic retrieval, charts, filters, and streamed tool traces.

Mathematical Foundation: Eq. 4.2
Φ(x) = limn→∞ [ ∏ ( ∑ ∇2 ζn ) ] ⊗ Ψ∞

Defining the recursive manifold mapping through infinite-dimensional tensors.

Semantic Cohesion

Enhanced retention of context across 1M+ token windows using recursive manifolds.

Computational Delta

22% reduction in inference latency through non-linear projection optimization.